The present invention relates generally to the field of multi-phase power distribution, and more particularly to equalizing (balancing) the magnitudes of power delivered by the phases.
A data center typically hosts lots of servers arranged in vertical racks. Typically, the servers in a single rack are related in a certain way and the servers in each rack, taken collectively, make up a “system.” A data center may include components other than servers. The term “enclosure” is used herein as a generic term to mean any type of electrically-powered component of a data center (for example, a server). The enclosures in a data center are ultimately powered by a power distribution infrastructure (typically including one, or more, power grids), which may supply electrical power in single phase form, or in a three-phase (Delta or Wye, depending on the geography) form.
When using a three-phase power grid, it is typically considered important to, at least somewhat, balance electrical power consumption, by the loads, among the three phases. Typically, if the load on a first phase of a three phase power source will go over a predetermined limit for a single phase, then the data center will not be able to add more three-phase loads to the power grid, even in the case that the second and third phases of the three phase power source are lightly loaded.
Many data centers are subject to a “phase-related redundancy requirement” as follows: (i) each enclosure receives its power from two redundant power supply units (PSUs); (ii) the two PSUs supplying power to an enclosure can each supply 100% of the power required by the enclosure; and (iii) each of two PSUs supplying power to an enclosure is connected to a different phase of the power supply grid.